1
Michael Schulze Linguistic Colloquium, 29/06/2021 PhD student Seminar für Afrikawissenschaften Humboldt-Universität zu Berlin Humboldt-Universität zu Berlin
Finding cognates in “South Atlantic”
1 Introduction
1.1 Background of this study PhD project: - historical-comparative reconstruction of the nominal classification systems in the Mel languages1 - dissertation embedded in the project “Noun classification systems in Africa between gender and nominal declension” at Humboldt Universität zu Berlin, funded by the “Deutsche Forschungsgemeinschaft” and headed by Tom Güldemann (see Güldemann 2016)
1.2 The Mel languages within “South Atlantic”
1.2.1 “South Atlantic” Countries: Sierra Leone, Guinea, Liberia (see Eberhard et al. 2021) Traditional classification (adapted from Sapir 1971: 49):
Niger-Congo Atlantic South Atlantic A. Sua B. Mel (including Gola) C. Limba
Figure 1: External and internal classification of “South Atlantic”
1 Funding of the dissertation project: ‘Elsa-Neumann-Stipendium des Landes Berlin’ and ‘Deutsche Universitätsstiftung (Gerda-Henkel-Stipendium)’
2
1.2.2 Mel - internal classification of Mel: two Branches (Northern Mel and Southern Mel), excluding Gola (adapted from Hammarström et al. 2021)
Northern Mel Baga Koba Baga Manduri Baga Sitemu Landuma Temne
Southern Mel Bullom Northern Bullom Bom-Kim (with Krim as a dialect) Bullom So (aka Mani, see Childs 2011) Sherbro Kisi Northern Kisi Southern Kisi
Figure 2: Internal classification of the Mel languages
1.3 Goals and methods of this talk - The identification of cognate nouns in Mel is an important task to disentangle the complex developments of the nominal classification systems. - recognition of cognates within each of the two branches mostly straightforward - however: limited number of obvious cognates between Northern and Southern Mel → application of the classical historical-comparative method on the basis of regular sound correspondences → presentation of some complex sound correspondences between the two branches in this talk → through application of classical method detection of larger number of cognates possible → strict application of the method also leads to the identification of cognates between Mel and the two single languages of “South Atlantic” Limba and Gola, which have not been obvious at first sight - What is presented in this talk is work in progress rather than sophisticated results. So, the reconstructions presented are preliminary and are intended to make clear what results can be expected in future work. 3
1.4 Language sample and data
1.4.1 Language sample: - Northern Mel: Temne, Baga Koba, Baga Sitemu, Landuma - Southern Mel: Mani, Krim, Sherbro, Kisi - Gola - Limba - no data for Sua included
1.4.2 Data - 80-word-lists of nouns (modified Leipzig-Jakarta list) - data extracted from dictionaries and lexical lists, some from grammatical descriptions (see Table 1) - data presented in the orthography used in the sources - reconstructed forms presented in IPA
Branch Language Data sources Northern Mel Temne Dalby 1966, Schlenker 1880, Bai-Sheka 1981, Kanu 2009, Kanu 2012, Kamarah 1994, Kamarah 2007, Wilson 2007, Wilson 1961, Yillah 1992 Baga Koba Relich 1973 Baga Sitemu Lamp 2016 Landuma Sumbatova 2012 Southern Mel Krim Pichl 1972, (Pichl 1976) Sherbro Pichl 1964 Mani Childs 2012, (Childs 2011) Kisi Childs 2000 Gola Westermann 1921 Limba Clarke 1922 Table 1: Data sources
4
1.5 Sapir’s lexicostatistical study - Sapir 1971: lexicostatistical study of Atlantic languages (based on Swadesh’s ‘First 100’) - results for “South Atlantic” (see Table 2): - high percentages within each branch of Mel (grey cells) - low percentages between Northern and Southern Mel (orange cells) - low percentages between Gola and Northern Mel, slightly higher percentages between Gola and Southern Mel (due to language contact?) - very low percentages between Limba and the other “South Atlantic” languages - very low percentages between Sua and the other “South Atlantic” languages (slightly higher between Sua and Northern Mel)
Sua Sua NM Temne 19 Temne Baga Koba 19 79 Baga Koba SM Mani 14 24 21 Mani Sherbro 13 21 22 82 Sherbro Krim 14 20 19 68 73 Krim Kisi 14 22 23 57 52 50 Kisi Gola 14 23 21 31 28 31 25 Gola Limba 13 14 16 13 12 11 10 14 Limba Table 2: Lexicostatistical study of “South Atlantic” (adapted from Sapir 1971: 47)
2 Cognate search based on regular sound correspondences
2.1 Coronal consonants
2.1.1 Two voiceless coronal plosives as a regional feature - distinction of two t-sounds in Temne (Northern Mel), the Bullom (Southern Mel) languages Krim and Sherbro as well as in Limba - absence of this distinction in non-Temne Northern Mel, but regular sound correspondences with each of the two sounds can be observed - no such distinction in the remaining Southern Mel languages Mani (Bullom language closely related to Krim and Sherbro) and Kisi - so far, no other languages (e.g., surrounding Mande languages) known to have this contrast - possibly, the distinction first evolved in Temne and then spread to neighboring languages - comparative study of the distinction in different Temne dialects desirable
5
Temne: - Wilson (1961: 3): distinction of the sounds Krim4 - Sherbro: - Limba: - voiceless alveolar plosive 2 Yillah writes [č] 3 Yillah writes /ṭ/ 4 Pichl’s orthography used here, no IPA 5 like English 2.1.2 Sound correspondences of coronal voiceless plosives in Northern Mel 2.1.2.1 Temne /t / as a reflex of Proto-Northern-Mel *C -Temne /t/ reflex of Proto-Northern-Mel *C (plosive with possibly palatal or lamino-palato- alveolar articulation) blood sound correspondence Temne mə̀- t ì r t Baga Koba ma- ts i r ts Baga Sitemu mə- ts i r ts Landuma mə- c i r c Proto-NM *ma- C i r *C Table 3: Reflexes of Proto-Northern-Mel *C - reflex can be [tʃ] in Baga Sitemu, e.g. a-tshen ‘dog’ 2.1.2.2 Temne /t/̪ as a reflex of Proto-Northern-Mel *t - Temne /t/̪ (Yillah 1992: /ʈ/) reflex of Proto-Northern-Mel *t (plosive with apico-dental or apico-alveolar articulation) (fire)wood sound correspondence Temne ɛ- t ̪ ɔ k6 t ̪ Baga Koba e- t o k t Baga Sitemu ku- t ɔ k t Landuma kə- t ɔ k t Proto-NM *t ɔ k7 *t Table 4: Reflexes of Proto-Northern-Mel *t 2.1.3 Reflexes of Proto-NM *C and *t in Southern Mel - Proto-Northern-Mel *C can be traced back to two different Proto-Mel consonants; in the first series, it corresponds to Proto-Southern-Mel *t (see Table 5), in the second to Proto-SM *s (see Table 6) - unclear, which Proto-Mel consonants have to be reconstructed in these cases - Proto-Northern-Mel *t originates in Proto-Mel *t (see Table 7) 6 Schlenker’s (1880: 398) original orthography: e-̱ toḵ ‘firewood’ 7 Where possible, prefixes are also reconstructed. 7 2.1.3.1 Proto-NM *C – Proto-SM *t dog sound correspondence Kisi citation form8 Temne ɑ̀- t ə̀ n t Baga Koba a- ts i n ts Baga Sitemu a- tsh e n tʃ Landuma a- c e n c Proto-NM *C V n *C Mani °ù- t ù m ɛ̀ t Krim th u n gbayɛ t ̪ Sherbro ø- th u m ɔɛ t ̪ Proto-Bullom t ̪ u m *t ̪ Kisi √ t ù ŋ t tùŋ=ndó Proto-SM *t u N *t Proto-Mel ? ? Table 5: Proto-NM *C vs. Proto-SM *t - Proto-Northern Mel *C could have originally been an allophone of *t before high vowels, so that *t would have to be reconstructed 2.1.3.2 Proto-NM *C – Proto SM *s dog sound correspondence Kisi citation form Temne ɑ̀- t ɔ́ ↓k ɔ́ t Baga Koba ta- ts o k o ts Baga Sitemu a- ts ɔ k ɔ ts Landuma a- c ɔ g ɔ c Proto-NM *C ɔ k ɔ *C Mani °ì- s ɔ̀ k s Krim s ɔ g s Sherbro ø- s ɔ k s Proto-Bullom *s ɔ k *s Kisi √ s ɔ̀ɔ́ s sɔ̀ɔ́ Proto-SM *s ɔ (k) *s Proto-Mel ? ? Table 6: Proto-NM *C – Proto SM *s 8 = root + class-speficific enclitic 8 - a reconstruction of *s for Proto-Mel is not probable because there is already a clear correspondence of Proto-Northern Mel *s and Proto-Southern-Mel *s (see Table 8); possibly, the sound to be reconstructed here is a Proto-Mel *ʃ 2.1.3.3 Proto NM *t – Proto-SM *t – Proto-Mel *t - series in Table 7 is the only one which has been found for this correspondence - therefore, these reconstructions are weak, also because a Temne reflex has not been found so far termite/flying ant sound correspondence Kisi citation form Temne -/- Baga Koba a- t e t ̪ Baga Sitemu a- t e t Landuma a- t e t Proto-NM *a t e *t Mani -/- Krim i- th ɛ t ̪ Sherbro °ø- th ɛ t ̪ Proto-Bullom *t ̪ ɛ *t ̪ Kisi √ t ɛ̀ t tìɔ́ Proto-SM *t ɛ *t Proto-Mel *t ɛ *t Table 7: Proto NM *t – Proto-SM *t – Proto-Mel *t 2.1.2 Sibilants in Mel - Mel has only voiceless sibilants, voiced sibilants are absent - an /ʃ/ phoneme is absent in all Mel languages - in Temne, the /s/ phoneme has the allophones [s] and [ʃ] (Yillah 1992: 15-16); they appear in free variation before front vowels, before back vowels, only [s] is possible - in Temne, an affricate phoneme /ʧ/ is absent (see Wilson 1961: 3 and Yillah 1992: 15) - in other Northern Mel languages, /ʦ/ or /ʧ/ exist as a sound correspondence of Temne /t/ (see 2.1.2.1) - in Southern Mel, /ʧ/ is present in all languages considered in this study 9 2.1.2.1 Proto-NM *s – Proto-SM *s – Proto-Mel *s - /s/ phoneme can be clearly reconstructed for Proto-Mel, having the reflex *s both in Proto- Northern-Mel and Proto-Southern Mel sand sound correspondence Kisi citation form Temne kə- s ə n t s Baga Koba Baga Sitemu kə- s ə n tsh s Landuma kə- s ə n c s Proto-NM *kə- s ə n C *s Mani °ù- s ú n s Krim Sherbro i- s û ng s Proto-Bullom *s u N *s Kisi √ s ìò ŋ s sìòŋ=ndé Proto-SM *s V N *s Proto-Mel *s V N *s Table 8: /s/ as a reflex of Proto-Mel *s throughout Mel - concluding from its absence in Proto-Southern-Mel, *C is most probably a frozen suffix in Proto-Northern-Mel; if this frozen suffix aligns with the prefix *kə-, in this case, *C can could be traced back to a *k which underwent palatalization 10 2.1.2.2 Proto-NM *s – Proto-SM *ʧ - Proto-Northern-Mel *s may correspond to Proto-Southern-Mel *ʧ (see Table 9, only this example found so far) - reconstruction for Proto-Mel unclear, possibly *ʧ - note that the reflexes of Northern-Mel in Table 9 all contain mid or high front vowels; thus, at least in Temne the phonetic realization can be [s] or [ʃ] (free variation according to Yillah (1992: 15)) tooth sound correspondence Kisi citation form Temne rə̀- s é k s Baga Koba da- s i k s Baga Sitemu de- s e k s Landuma da- s e g s Proto-NM *da- s e k *s Mani ì- c á ŋ ʧ Krim c a ng ʧ Sherbro ø- c a ŋ ʧ Proto-Bullom *ʧ a ŋ *ʧ Kisi √ c ì ŋ ʧ cìŋ-ndé Proto-SM *ʧ V ŋ *ʧ Proto-Mel ? ? Table 9: Proto-NM *s – Proto-SM */ʧ/ 2.1.2.3 Proto-NM syllable-final *s – Proto-SM *l - interestingly, syllable-final *s in Proto-Northern Mel patterns with *l in Proto- Southern Mel - in Table 10, *l is tentatively reconstructed for Proto-Mel, considering that distantly related Limba has the cognate form fēli ‘egg’ - on the other hand, both in Mel and in Limba a voiced sibilant /z/ is absent so that it cannot be excluded that indeed a form *fez has to be reconstructed - Tables 11 and 12 show that the Northern Mel forms (if really cognate in these cases) can be the result of the loss of a final vowel - in Table 12, the Kisi correspondence /d/ gives a hint, that *d rather than *l has to be reconstructed - sound shift *l > *s in Northern Mel obscures the cognacy of the respective lexemes between the two branches of Mel so that a lexicostatistic study may fail to detect such cognates (indeed, Sapir (1971) did not recognize the case presented in Table 10) 11 egg sound correspondence Kisi citation form Temne rə̀- m ɛ̀ s s Baga Koba da- m e s s Baga Sitemu a- m ɛ s s Landuma da- m ə s s Proto-NM *da- m E s *s Mani °ù- p è l l Krim p ui n -tha n Sherbro ø- p e l l Proto-Bullom p e l *l Kisi √ p è l l pèl=téŋ Proto-SM *p e l *l Proto-Mel *p e l *l Table 10: Syllable-final *s in Proto-NM – *l in Proto SM (1) star sound correspondence Kisi citation form Temne k-ɔ̀ s s Baga Koba k-o s s Baga Sitemu k-ɔ s s Landuma k-ɔ s s Proto-NM *k-ɔ s *s Mani °ù- l è l Krim l ê l Sherbro ø- l e l Proto-Bullom *l e *l Kisi ?√ l ò l lùé Proto-SM *l V *l Proto-Mel *l V *l Table 11: Syllable-final *s in Proto-NM – *l in Proto SM (2) 12 name sound correspondence Kisi citation form Temne ŋ-é s s Baga Koba Baga Sitemu Landuma Proto-NM Mani °ì- l é l ì l Krim i- l e n l Sherbro i- l e l l Proto-Bullom *l e l *l Kisi √ d ìò d dìò=láŋ Proto-SM *l V *l or *d Proto-Mel *l V *l or *d Table 12: Syllable-final *s in Proto-NM – k *l in Proto-SM (3) 2.2 Non-coronal consonants 2.2.1 Labial obstruents 2.2.1.1 Reconstruction of Proto-Mel *b and *f - Proto-Mel *b and *f can be reconstructed straightforwardly for some nouns (Tables 13 and 14) head sound correspondence Kisi citation form Temne rə̀- b ó ŋ p b Baga Koba da- b u m p b Baga Sitemu do- b o m p b Landuma da u m p Proto-NM *da- b U m p *b Mani °ù- b ó l b Krim b u n b Sherbro ø- b o ̹ l b Proto-Bullom *b o l *b Kisi √ b ò l b bòl=léŋ Proto-SM *b o l *b Proto-Mel *b o l *b Table 13: Proto-NM *b – Proto-SM *b – Proto-Mel *b 13 - Limba hu-ya ‘head’ and Gola má-dî ‘head’ are probably cognate and support the reconstruction of the final /l/ in Proto-Mel *bol; moreover, they point out, that *bo may be a frozen prefix in Proto-Mel - final /p/ in Proto-Northern Mel is most probably a frozen suffix which possibly aligns with the frozen prefix *bo of Proto-Mel tooth sound correspondence Kisi citation form Temne rə̀- f ɔ̀ r f Baga Koba da- f o r f Baga Sitemu dɔ- f ɔ r f Landuma da- f ɔ r f Proto-NM *da- f ɔ r *f Mani °dì- f ɔ̀ l f Krim tha h u n h Sherbro ø- h ɔ l h Proto-Bullom *f ɔ l *f Kisi √ h ɔ̀ l h hɔ̀l=téŋ Proto-SM *f ɔ l *f Proto-Mel *f ɔ r *f Table 14: Proto-NM *f – Proto-SM *f – Proto-Mel *f - unclear, whether in Proto-Mel final *r or *l has to be reconstructed - if Proto-Mel has *l, the question arises, why there is no final /s/ in Northern Mel, as it is the case in the examples in 2.1.2.3; maybe the quality of a former final vowel (e.g., high vs. non-high) plays a role - Limba fo-ya ‘eye’ and Gola é-fe ̱̀ ‘eye’ support the reconstruction of the above series 14 2.2.1.2 Proto-NM *f – Proto-SM *p – Proto-Mel *p - if Proto-Northern-Mel *f corresponds to Proto-Southern-Mel *p, *p is reconstructed for Proto- Mel - not completely sure whether the Kisi form in Table 15 belongs to this series person/child sound Kisi citation form and correspondence semantic remarks Temne à- f ə̂ m f ‘person' (PL) Baga Koba i- f o m f ‘person' Baga Sitemu ø- f u m f ‘person' Landuma ø- f u m f ‘person' Proto-NM *f u m *f Mani °ù- p ɔ̀ m p 'child' Krim ka- p u m a p 'son' Sherbro ø- p u m ɔ p 'child (boy or girl)' Proto-Bullom *p u m V *p Kisi √ p ìà nd ù p pìàndòó 'male' Proto-SM *p u m V *p Proto-Mel *p u m V *p Table 15: Proto-NM *f – Proto-SM *p – Proto-Mel *p (1) - in the series shown in Tables 16 and 17, in Northern Mel the /f/ is word-final, whereas in Southern-Mel the /p/ is followed by VC (VCV in Kisi) - the cognacy becomes visible through reconstruction, but is not immediately apparent in a lexicostatistic comparison - the Southern Mel, the VC sequence can be possibly traced back to a sequence of a final root vowel and a following petrified or still productive suffix 15 earth, soil, country sound correspondence Temne ɑ̀-thɔ̂ f f Baga Koba Baga Sitemu a-tɔ f f Landuma a-tɔ f f Proto-NM a-tɔ *f *f Mani Krim °ù- p ɔ̀ k p Sherbro ku- p ɔ g p Proto-Bullom ø- *p ɔ k *p Kisi Proto-SM Proto-Mel *p V C *p Table 16: Proto-NM *f – Proto-SM *p – Proto-Mel *p (2) moon, month sound Kisi citation + prefixed correspondence form Temne ŋ-ó f f Baga Koba n-wo f f Baga Sitemu ŋ-o f f Landuma ŋ-o f f Proto-NM *ŋ-o f *f Mani °ì- p à n p Krim i- p a ŋ p Sherbro i- p a ŋ p Proto-Bullom *i- p a ŋ *p Kisi √ p à ŋ ù p pàŋéí; prefixed form: ì-pàŋù Proto-SM *p a ŋ (V) *p Proto-Mel *p V C (V) *p Table 17: Proto-NM *f – Proto-SM *p – Proto-Mel *p (3) - probably cognate form in Limba: seṅkele ‘moon’; seṅ could be petrified prefix material, kele the part of the word form which is cognate to Proto-Mel *pVC(V) - if this conclusion is correct, Proto-Mel *p corresponds to Limba /k/ - another hint at the correspondence p – k is the probable cognacy of Proto-Southern-Mel *di-pal ‘sun’ and Limba kaṅ ‘sun’ - the corresponding Gola lexemes are ó-wàla ‘moon’ and é-gwee̱̍ ̱́ ‘sun’, both appear to be cognate with the Mel and Limba lexemes 16 2.2.1.3 Proto-NM *b – Proto-SM *p - clear evidence of a correspondence Proto-Northern-Mel *b and Proto-Southern-Mel *p - unclear which plosive has to be reconstructed - the probably cognate Gola form ké-gòa (or ké-gwà) ‘bone’ points out that the plosive to be reconstructed for Proto-Mel may be a labiovelar, possibly *gb or *gw - the origin of final *t of Proto-Northern-Mel is unclear; the preceding nasal of Proto-NM and the final *k of Proto-Bullom look like petrified suffixes bone sound correspondence Kisi citation form Temne kə̀- b ɑ́ ŋ th b Baga Koba kə- b a n t b Baga Sitemu kə- b ɛ n t b Landuma kə- b a n t b Proto-NM *kə- *b V N t *b Mani °ù- p á k p Krim p a g p Sherbro ø- p a k p Proto-Bullom *p a k *p Kisi √ p àá p pàá Proto-SM *p a (k) *p Proto-Mel ? ? Table 18: Proto-NM *b – Proto-SM *p 2.2.1.3 Proto-NM *b – Proto-SM *w bird sound correspondence Kisi citation form Temne ɑ̀- b ɑ̂ ŋ p b Baga Koba a- b a m p b Baga Sitemu a- b ɛ m p b Landuma a- b a m p b Proto-NM *b V N p *b Mani °ù- w é w Krim v ɨ p v Sherbro ø- v ê v Proto-Bullom *w V p *w Kisi √ y ò ù y yòù=wó Proto-SM *w V p *w Proto-Mel ? ? Table 19: Proto-NM *b – Proto-SM *w 17 - possibly, *w has to be reconstructed or Proto-Mel, but with uncertainty - the obviously cognate Gola form ó-wê ‘bird’ supports this reconstruction, but the possibly cognate Limba form is bec ‘bird’ 2.2.2 Velar and labiovelar consonants - all Mel languages have labiovelar plosives - in the 80-word-lists, only few words with initial labiovelar plosives are attested, so that for reconstructional work more data has to be included - the labiovelars attested in the present database are found in Southern Mel (see especially Table 22); Northern Mel labiovelars are not attested 2.2.2.1 Proto-NM *k – Proto-SM *k – Proto-Mel *k - in Temne, the /k/ is part of a prefix, this could also be the case for the Proto-Mel form - the final k(V) sequence reconstructed for Proto-Mel could be a petrified or still productive suffix - possibly, in Proto Mel the root was *lu, marked by a circumfix *kV…k(V) seed sound correspondence Kisi citation form Temne °k ɨ̀- l ù k à k Baga Koba Baga Sitemu Landuma Proto-NM Mani Krim Sherbro k ɔ l u ng k Proto-Bullom Kisi √ k ú l ú k kúlóó Proto-SM *k V l u (ŋ) *k Proto-Mel *k V l u k (V) *k Table 20: Proto-NM *k – Proto-SM *k – Proto-Mel *k (1) - the following series shows some similarity with the previous one, but Krim (unattested in Table 20) has labiovelar /gb/, while Kisi has palatal /c/ - the labiovelar in Krim and the palatal in Kisi probably go back to Proto-Southern-Mel *k, the Kisi reflex being the result of palatalization before a high front vowel - *ki in Proto-Southern Mel and Proto-Mel is possibly a petrified or still productive prefix 18 house sound correspondence Kisi citation form Temne Baga Koba Baga Sitemu k ə- l ɔ k Landuma k ə- l ɔ k Proto-NM *k ə- l ɔ *k Mani °ù- k ì l k Krim gb â l ɛ gb Sherbro ø- k ī l k Proto-Bullom *k i l V *k Kisi √ c ìɛ̀ ì c cìɛ̀ì=yó Proto-SM *k i l V *k Proto-Mel *k i l V *k Table 21: Proto-NM *k – Proto-SM *k – Proto-Mel *k (2) 2.2.2.1 Proto-NM *k – labiovelar plosive in Proto-SM - Proto-Northern-Mel has *k, at least for Proto-Bullom and Proto-Southern-Mel, possibly also for Proto-Mel, a labiovelar plosive *KP is tentatively reconstructed, but it is hard to determine, whether it is *gb, *kp, *kw or even *gw - Gola has the cognate form ké-kombò ‘smoke’ - considering the Gola form and the series shown in Table 20, it is well possible, that Proto- Mel had *k in this series rather than a labiovelar plosive - prefixal initial /k/ of Northern Mel could be due to the reanalysis of a part of the root smoke (of fire) sound Kisi citation correspondence form Temne k- i m ä k Baga Koba k- i m a k Baga Sitemu k i- n i m ɛ k Landuma k- i m a n k Proto-NM *k i m V k Mani gb ì m í gb Krim yi- kw ɔ m kw Sherbro i- gb ĩ m i gb Proto-Bullom *KP V m (V) *KP Kisi √ k ǔ m k kǔm-ndé Proto-SM *KP V m (V) *KP Proto-Mel *KP V m V *KP Table 22: Proto-NM *k – labiovelar in Proto-SM 19 2.2.2.2 Proto-NM *k – Proto-SM *p – Proto-Mel *KP - series shown in Table 23 most probably goes back to a labiovelar plosive *KP (voiceless or voiced) in Proto-Mel - *KP developed into *k in Proto-Northern-Mel and *p in Proto-Southern-Mel - Limba ku-tagba ‘skin, hide, paper’ is possibly cognate and confirms the above reconstruction skin, hide, paper sound Kisi citation form and correspondence semantic remarks Temne ä- r e k a k 'skin, hide' Baga Koba a- r e k a k 'skin' Baga Sitemu a- r e k a k 'skin, paper' Landuma a- r e g a g 'skin, hide' Proto-NM *a- r e k a *k Mani °ù- r à ‘1) skin (animal) 2) cardboard’ Krim y a p p 'paper, book, letter' Sherbro ø- r æ 'book, paper' Proto-Bullom *r a p *p Kisi √ y à ú u yàú=wó ‘skin, hide, paper, letter, book’ Proto-SM *r a p *p Proto-Mel *r a KP (V) *KP Table 23: Proto-NM *k – Proto-SM *p – Proto-Mel *KP 20 3 Conclusions and outlook 3.1 Conclusions - The two voiceless coronal plosives /t/ and /t/̪ of Temne can be traced back to *C and *t in Proto-Northern Mel respectively, the former proto-sound most probably having been a palato- alveolar affricate /ʧ/ or a palatal plosive /c/. - This clearly shows that the distinction of /t/ and /t/̪ originated in Temne and spread through language contact to the neighboring languages Limba, Krim and Sherbro. - While Proto-Northern-Mel *t patterns with Proto-Southern-Mel *t and can thus be traced back to Proto-Mel *t, Proto-Northern-Mel *C obviously goes back to two different sounds in Proto-Mel which have to be determined yet. - The situation with sibilants is complex as well. The existence of palato-alveolar sibilants besides *s, possibly even *z, in Proto-Mel is probable. - Quite radical changes in Proto-Northern-Mel are responsible for the fact, that a number of cognates between Northern and Southern are hard to detect at first sight. The changes in Proto-Northern-Mel are: - Proto-Mel word-final *l(V) → Proto-NM word-final *s - Proto-Mel word-final CV(CV) → Proto-NM word-final C - It is possible that Proto-Mel had in both cases a monosyllabic CV stem which ended up as a word-final consonant in Proto-Northern-Mel. - This clearly shows the importance of the establishment of regular sound correspondences within the framework of the classical historical method. Otherwise, the identification of cognates that have undergone complex sound changes is hardly possible. The same can be said for many cognates between Mel and Limba or Mel and Gola which are difficult to recognize at first sight. Some examples have been given in this talk. 3.2 Outlook - Sound correspondences will be put on a more solid base by using a more extensive wordlist - A revision of Sapir’s lexicostatistic study (1971) including the cognates yielded by historical- comparative reconstruction would be interesting. - In the course of the historical-comparative reconstruction work, petrified affixes have been found in both branches of Mel. While both prefixes and suffixes are present in all Southern Mel languages, Northern Mel languages are strictly prefixing. So, petrified suffixes in Northern Mel are an especially interesting finding. It is my present hypothesis that Proto-Mel was strictly prefixing, the suffixes and circumfixes being an innovation in Southern Mel. If it can be shown that the suffixes which are nowadays petrified in Northern Mel were still active in Proto-Mel, this hypothesis has to be completely revised. 21 Abbreviations NM Northern Mel SM Southern Mel References Bai-Sheka, Abou. 1981. Le système nominal du temne. Grenoble: Université de Grenoble III, PhD thesis. Berry, Jack. 1958. Nominal classes in Hu-Limba. Sierra Leone Studies 10. 169-173. Childs, George Tucker. 2000. A dictionary of the Kisi language, with an English-Kisi index (Westafrikanische Studien: Frankfurter Beiträge zur Sprach- und Kulturgeschichte 22). Köln: Rüdiger Köppe. Childs, George Tucker. 2011. A grammar of Mani (Mouton Grammar Library 54). Berlin & Boston: De Gruyter Mouton. Childs, George Tucker. 2012. Mani lexicon. Unpublished. Clarke, Mary Lane. 1971 [1922]. A Limba-English dictionary/Tampeṅ Ta Ka Taluṅ Ta Ka Hulimba In Huiṅkilisi Ha. Westmead und Farnborough: Gregg International Publishers Limited. Dalby, David. 1966. Lexical analysis in Temne with an illustrative word list. Journal of West African Languages 3(2). 5-26. Eberhard, David M., Gary F. Simons, and Charles D. Fennig (eds.). 2021. Ethnologue: Languages of the World. Twenty-fourth edition. Dallas, Texas: SIL International. [Online version: http://www.ethnologue.com, last access 17/06/2021] Güldemann, Tom. 2016. Noun classification systems in Africa between gender and nominal declension. Project description – project proposals. Application to the DFG (Deutsche Forschungsgemeinschaft). Hammarström, Harald, Robert Forkel, Martin Haspelmath and Sebastian Bank. 2021. Glottolog 4.4. Leipzig: Max Planck Institute for Evolutionary Anthropology. https://doi.org/10.5281/zenodo.4761960 (Available online at http://glottolog.org, accessed on 2021-06-17.) Kamarah, Sheikh Umarr. 1994. Phonology and Tonology in the Morphology of Temne: A Lexicalist Approach. Ann Arbor: UMI, PhD thesis. Kamarah, Sheikh Umarr. 2007. A descriptive grammar of KʌThemnɛ (Temne) (Languages of the World/Materials 465). München: Lincom Europa. Kanu, Sullay Mohamed. 2009. Suffix ordering in Temne: a case for morphotactics. In Masangu Matondo, Fiona McLaughlin and Eric Potsdam (eds.), Selected proceedings of the 38th annual conference on African linguistics: linguistic theory and African language documentation, 141-150. Somerville, MA: Cascadilla Proceedings Project. Kanu, Sullay Mohamed. 2012. Valence-increasing morphology in Temne. Edmonton: University of Alberta, PhD thesis. 22 Lamp, Frederick John. 2016. Baga Tshi-Tem dictionary (Guinea) (revised May 2016). Yale University, unpublished ms. (ca. 3000 words). Pichl, Walter J. 1964. Sherbro-English dictionary. Freetown: Fourah Bay College. Pichl, Walter J. 1972. The Krim language in Sierra Leone. Pittsburgh: Duquesne University. Pichl, Walter J. 1976. Krim. In M. E. Kropp-Dakubu (ed.), West African Language Data Sheets: Volume 1. West African Linguistics Society. KRI 1-10. Relich, Randa Rae. 1973. General grammar of Koba Baga with texts and translations. Pittsburgh: Duquesne University, MA thesis. Sapir, J. David. 1971. West Atlantic: an inventory of the languages, their noun class systems and consonant alternation. In T. A. Sebeok (eds.), Current Trends in Linguistics 7, 45-112. Den Haag & Paris: Mouton. Schlenker, [Reverend] Christian Frederick. 1880. An English-Temne dictionary. London: Church Missionary Society. Sumbatova, Nina R. 2012. Materialy k slovaryu i teksty landuma. In Valentin F. Vydrin und A. Yu. Želtov (Hrsg.), Meždu Nigerom i Kongo: zametki na polyax. K 60-letiyu Konstantina Pozdnyakova, 170-230. Sankt-Petersburg: Nestor-Istoria. Westermann, Diedrich. 1921. Die Gola-Sprache in Liberia: Grammatik, Texte, Wörterbuch (Abhandlungen aus dem Gebiet der Auslandskunde 6). Hamburg: L. Friederichsen & Co. Wilson, William André Auquier. 1961. An outline of the Temne language. London: School of Oriental and African Studies (SOAS), University of London. Wilson, William André Auquier. 2007. Atlantic Wordlist based on Swadesh “First 100”. In Guinea Languages of the Atlantic group: description and internal classification (Research in African Studies 12), 197-244. Frankfurt am Main: Lang. Yillah, Mohamed Sorie. 1992. Temne phonology and morphology. New York: City University of New York, PhD thesis , the latter being a voiceless interdental plosive (no information on phoneme status) - Yillah (1992): phonemic distinction of two t-sounds: - /t/ with the allophones [t] and [ʧ]2; both allophones in free variation before front vowel, only [t] before back vowel (Yillah 1992: 16-17); [t] is laminal and aspirated (Yillah 1992: 13) -/ʈ/3 (written ) as an apical retroflex stop without allophones (Yillah 1992: 12- 13, 18-19) - according to Yillah (1992: 18-19) difference in the manner of articulation of both sounds, possibly /ʈ/ bearing the feature [+RTR] with a “dampened” quality and lower pitch of the following vowel, whereas this feature is absent on /t/ as distinct phonemes (Pichl 1972: 1) - , main difference is a stronger aspiration of always dental, in initial position slightly aspirated (Pichl 1972: 2) like dental ‘t’, but not English ‘θ’ or ‘ð’ (Pichl 1964: X) - no information on phonemic status of the two sounds ; no information on phonemic status (Berry 1958: 169) - no distinction of the two sounds in Clarke’s dictionary (1922), invariably